Chat Completions

Create Chat Completion

POST/v1/chat/completions

Generate a text completion.

---

Request

curl -X POST https://api.ig1.ai/v1/chat/completions \
  -H "Authorization: Bearer $IG1AI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k26-instant",
    "messages": [
      { "role": "user", "content": "Write me a nice poem about DevOps." }
    ]
  }'

Request body

Parameter	Type	Required	Description
model	string	✅	Model ID (see LLM Models)
messages	array	✅	Array of message objects (role + content)
stream	boolean		Enable streaming (true/false)
temperature	float		Sampling temperature
top_p	float		Nucleus sampling
max_tokens	integer		Maximum tokens to generate

Extra sampling parameters

The following additional sampling parameters are supported:

Parameter	Type	Description
use_beam_search	bool	Enables beam search over sampling.
top_k	int	Limits sampling to top K tokens (1 to Vocabulary_size).
min_p	float	Minimum probability relative to top token (0.0 to 1.0).
repetition_penalty	float	Penalizes repeated tokens (from 1.0).
length_penalty	float	Adjusts sequence length in beam search (from 0.0 to 2.0+).
stop_token_ids	list[int]	Tokens that trigger generation end (e.g., []).
include_stop_str_in_output	bool	Keeps stop strings in final text.
ignore_eos	bool	Forces generation past EOS token.
min_tokens	int	Minimum tokens to generate (0 to max_tokens).
skip_special_tokens	bool	Removes special tokens from output.
spaces_between_special_tokens	bool	Adds spacing between special tokens.
truncate_prompt_tokens	int	Limits prompt length by truncating (from 1 to max_context).
prompt_logprobs	int	Returns logprobs for prompt tokens (e.g., null).
allowed_token_ids	list[int]	Restricts output to specific tokens (typical values: 0 to 5).
bad_words	list[string]	Prohibits specific words in output.

---

Response

{
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "logprobs": null,
      "message": {
        "annotations": null,
        "audio": null,
        "content": " **The Infinite Loop**\n\nIn the space where code meets the breathing machine ...",
        "function_call": null,
        "reasoning": null,
        "refusal": null,
        "role": "assistant",
        "tool_calls": []
      },
      "routed_experts": null,
      "stop_reason": 163586,
      "token_ids": null
    }
  ],
  "created": 1780560755,
  "id": "chatcmpl-82a17a861416606b",
  "kv_transfer_params": null,
  "model": "kimi-k26-instant",
  "object": "chat.completion",
  "prompt_logprobs": null,
  "prompt_text": null,
  "prompt_token_ids": null,
  "service_tier": null,
  "system_fingerprint": "vllm-0.22.0-tp8-a8ec695b",
  "usage": {
    "completion_tokens": 276,
    "prompt_tokens": 18,
    "prompt_tokens_details": null,
    "total_tokens": 294
  }
}

---

Unsupported OpenAI endpoints

warning

IG1 AI does not store any chat history or data. That is why following endpoints are not supported:

• GET/v1/chat/completions
• GET/v1/chat/completions/{id}
• POST/v1/chat/completions/{id}
• DELETE/v1/chat/completions/{id}